INDEX

Explanations

appropriations and hand gestures

The neuron flags content-bearing (i.e. non–stopword) tokens rather than common function words.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

何故

-1.52

 these

-1.37

Because

-1.36

 subsequent

-1.34

<bos>

-1.31

揸

-1.31

薆

-1.25

各種

-1.23

when

-1.23

 protege

-1.23

POSITIVE LOGITS

籥

1.52

 majeure

1.42

งิน

1.40

1.38

JAJA

1.36

thansa

1.34

高端

1.31

entrion

1.29

槪

1.28

 capuche

1.27

Activations Density 0.003%